Here at Eigen, we first came across this category of problems back in 2016.
It all started with the problem of knowing which maintenance backlog items to solve for the biggest impact on the safety of the facility. Different maintenance tickets affected different safety and environmental critical elements (SECEs) differently and different SECEs are linked in a bowtie diagram of barriers, so the priority should be given to the bowties diagram paths that are most impaired a result of SECEs that affect the most barriers.
The solution was the development and implementation of the Barrier Safety Panel (BSP): When first implemented, the users were able to reduce the Safety Critical backlog by 90% in 3 weeks as a result of the prioritisation focus provided by the BSP model[JC1] .
The BSP model used a Knowledge Graph: A mathematical description of nodes and their relationships. In the BSP model, different maintenance backlog conditions (preventive or corrective, overdue, ICSS detected issues, sensor inhibits) were assigned a different score and then the scores were aggregated using the graph relationships to calculate the relative level of impairment at every level in the barrier architecture. Organisational impairments (Drills, Training Status and Audits) can also be added to the model.
Then came the problem of Asset Upset Risk: It was posited that a service (Gas Compression, Water Injection, etc.) within a large multi-asset offshore oil field had a higher risk of shutdown depending on different conditions such as maintenance backlog size or overdue condition, identified risks or number of active alarms, to name a few factors. So quite simply it can be inferred that the risk of shutdown is proportional to the number of conditions identified, but in reality, systems and services are all interacting with each other so the loss of one service (e.g. Electrical supply) can affect the availability of other services (e.g. Pumps), all of which can be modelled using a Knowledge Graph too. Determining asset with the highest risk of upsets was important to focus the operational and engineering teams’ attention and avert shutdowns, which can cascade to the entire operation.
Another customer was interested in focusing their attention on the Prioritisation of corrosion condition inspections to be done in a large facility: The human resources dedicated to corrosions inspections is limited and the facility is large and distributed over a large geographical area, and therefore a solution to prioritise efforts was required: Which points in the process have the highest corrosion-related risk? Which are redundant or pose a relatively low risk? Once again, a graph describing the structure of the plant and factors determining more or less risk, such as location (also in relation to other inspection points), exposure to corrosion and tubular shape would provide valuable insights.
Knowledge Graphs play a valuable role in solving all these problems: At Eigen we use open-source Neo4j as the graph technology embedded in our solutions, so access to the graph describing your problem is accessible via an open API and not locked-in a closed monolithic solution.
We have been implementing practical decision support systems based on Knowledge Graphs since 2016. The decision support system connects online to functional systems such as CMMS, Risk Registers and Control System to periodically extract “signals” that can then be aggregated and summarised using the business rules embedded in the Graph and displayed in the standard or customised front-end displays, always providing the latest status and identifying clear priorities for the maintenance or operational staff.
For more information about how we can solve your decision-making problem, get in touch.